{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 22080,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 2.9320652173913043e-05,
      "loss": 2.2272,
      "step": 500
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.8641304347826088e-05,
      "loss": 1.6649,
      "step": 1000
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.796195652173913e-05,
      "loss": 1.5016,
      "step": 1500
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.7282608695652175e-05,
      "loss": 1.4064,
      "step": 2000
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.6603260869565217e-05,
      "loss": 1.3953,
      "step": 2500
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.592391304347826e-05,
      "loss": 1.3777,
      "step": 3000
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.5244565217391304e-05,
      "loss": 1.2818,
      "step": 3500
    },
    {
      "epoch": 0.36,
      "learning_rate": 2.456521739130435e-05,
      "loss": 1.2598,
      "step": 4000
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.3885869565217394e-05,
      "loss": 1.2099,
      "step": 4500
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.3206521739130436e-05,
      "loss": 1.1691,
      "step": 5000
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.2527173913043478e-05,
      "loss": 1.1772,
      "step": 5500
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.1847826086956523e-05,
      "loss": 1.1432,
      "step": 6000
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.1168478260869565e-05,
      "loss": 1.1305,
      "step": 6500
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.048913043478261e-05,
      "loss": 1.1243,
      "step": 7000
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9809782608695652e-05,
      "loss": 1.1183,
      "step": 7500
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.9130434782608694e-05,
      "loss": 1.0926,
      "step": 8000
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.845108695652174e-05,
      "loss": 1.0649,
      "step": 8500
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.777173913043478e-05,
      "loss": 1.0535,
      "step": 9000
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.7092391304347826e-05,
      "loss": 1.0591,
      "step": 9500
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6413043478260868e-05,
      "loss": 1.0451,
      "step": 10000
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.5733695652173913e-05,
      "loss": 1.0146,
      "step": 10500
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5054347826086957e-05,
      "loss": 1.0105,
      "step": 11000
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.4375e-05,
      "loss": 0.7476,
      "step": 11500
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.3695652173913042e-05,
      "loss": 0.7657,
      "step": 12000
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.3016304347826088e-05,
      "loss": 0.7341,
      "step": 12500
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.2336956521739131e-05,
      "loss": 0.7278,
      "step": 13000
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.1657608695652175e-05,
      "loss": 0.7484,
      "step": 13500
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.0978260869565218e-05,
      "loss": 0.6922,
      "step": 14000
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.0298913043478262e-05,
      "loss": 0.7404,
      "step": 14500
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.619565217391304e-06,
      "loss": 0.7461,
      "step": 15000
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.940217391304347e-06,
      "loss": 0.7499,
      "step": 15500
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.260869565217392e-06,
      "loss": 0.7297,
      "step": 16000
    },
    {
      "epoch": 1.49,
      "learning_rate": 7.581521739130435e-06,
      "loss": 0.7272,
      "step": 16500
    },
    {
      "epoch": 1.54,
      "learning_rate": 6.9021739130434785e-06,
      "loss": 0.7145,
      "step": 17000
    },
    {
      "epoch": 1.59,
      "learning_rate": 6.222826086956522e-06,
      "loss": 0.6961,
      "step": 17500
    },
    {
      "epoch": 1.63,
      "learning_rate": 5.543478260869565e-06,
      "loss": 0.69,
      "step": 18000
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.864130434782609e-06,
      "loss": 0.7141,
      "step": 18500
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.184782608695653e-06,
      "loss": 0.7009,
      "step": 19000
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.5054347826086958e-06,
      "loss": 0.6956,
      "step": 19500
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.8260869565217393e-06,
      "loss": 0.6821,
      "step": 20000
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.1467391304347824e-06,
      "loss": 0.6994,
      "step": 20500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.4673913043478262e-06,
      "loss": 0.6749,
      "step": 21000
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.880434782608695e-07,
      "loss": 0.6758,
      "step": 21500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.0869565217391305e-07,
      "loss": 0.6613,
      "step": 22000
    },
    {
      "epoch": 2.0,
      "step": 22080,
      "total_flos": 5.192058010209178e+16,
      "train_loss": 0.9816198554591856,
      "train_runtime": 6662.5688,
      "train_samples_per_second": 39.765,
      "train_steps_per_second": 3.314
    }
  ],
  "max_steps": 22080,
  "num_train_epochs": 2,
  "total_flos": 5.192058010209178e+16,
  "trial_name": null,
  "trial_params": null
}